The Journal of
Physical Chemistry
@3 Copyright, 1991, by the American Chemical Society
VOLUME 95, NUMBER 2 JANUARY 24,1991
LETTERS Imaging of Colloidal Goid on Graphite by Scanning Tunneling Microscopy: Isolated Particles, Aggregates, and Ordered Arrays J. F. Womelsdorf,* W. C . Ermler,* Department of Chemistry and Chemical Engineering and Department of Physics and Engineering Physics, Stevens Institute of Technology, Hoboken, New Jersey 07030
and C. J. SandrofP Bellcore, Red Bank, New Jersey 07701 (Received: August 7 , 1990; In Final Form: November 16, 1990)
Gold colloidal suspensions have been deposited onto highly oriented pyrrolitic graphite and analyzed by using a scanning tunneling microscope (STM). The size distribution of individual particles measured with the STM (I6 1.5 nm) agreed well with measurements made by transmission electron microscopy. STM images of colloidal aggregates containing up to 15 individual particles could be obtained under a wide variety of experimental conditions. The quality of these images was extremely sensitive to the type of aggregating agent used, with ionic salts giving better results than molecular species. Under our colloidal deposition conditions we often observed highly ordered hexagonal patterns with a periodicity of 75-1 SO nm, extending over areas as large as 150 X 150 nm2. Preliminary evidence suggests that this hexagonal lattice is associated with the graphite substrate itself, and not with an ordered, two-dimensional array of colloidal particles.
*
Perhaps the most significant advantage of the scanning tunneling microscope (STM) over other surface science tools is its ability to scan large areas of a surface and to probe local, nonperiodic structures with atomic scale resolution. This feature is of great utility in searching for adsorbed chemical species ranging from molecules' to metallic and semiconductor clusters.2-6 ( I ) Hallmark, V. M.; Chiang, S.; Rabolt, J. F.; Swalen, J. D.; Wilson, R. J. Phys. Rev.Lett. 1987. 59, 2819. (2) Abraham, D. W.; Sayyler, K.; Ganz. E.; Mamin, H. J.; Thomson, R. E.; Clarke, J. J . Appl. Phys. Lett. 1986, 49, 853. (3) Ganz, E.; K. Sattler, K.; Clarke, J. J . Yac. Sci. Technol. 1988, Ab, 419. (4) Baro, A. M.;Bartolome, A.; Vazquez, L.; Garcia, N.; Reifenberger, R.; Choi, E.; Andres, R. P. Appl. Phys. Lett. 1987, 51, 1594. ( 5 ) Sarid, D.;Henson, T.; Bell, L. S.; Sandroff, C. J. J . Vac. Sci. Technol. 1988, Ab, 426.
0022-3654/91/2095-0503$02.50/0
Metallic clusters adsorbed on graphite have had a special appeal for STM workers24 because they are easily produced, are relatively robust, and are readily differentiated from the graphite substrate through their tunneling characteristics. Encouraged by the STM results on small metal clusters, we began to explore the properties of larger metallic systemscolloidal particles-deposited on the surface of highly ordered pyrolytic graphite (HOPG).In principle, many important features of colloids could be understood by exploiting the atomic resolution achievable with the STM. For example, local variation of the electrical double layer surrounding colloidal particles, and the geometry of adsorbed molecular species on colloidal surfaces, are (6) Jing, T. W.; Ong,N. P.; Sandroff, C. J. Appl. Phys. Lett. 1988,53, 104.
0 1991 American Chemical Society
504 The Journal of Physical Chemistry, Vol. 95, No. 2, 1991
Letters the presence of conductive metallic particles strongly adsorbed on the HOPG substrate. To confirm that these images were of gold colloids and not random adsorbed impurities or graphite artifaces, we measured the dimensions of the particles in the xy direction (parallel to the graphite substrate) with the STM. This measurement gave an average particle diameter of 16 f 1.5 nm. Considering that tip-cluster interactions can change the apparent size of measured features," this diameter is excellent agreement with the 14.5-nm colloid diameter measured by TEM in this and in previous work.* Though xy scans with the STM accurately reflected gold particle diameters, we found that the height mode was not quantitatively reliable. Generally, the STM gave particle heights in the z direction varying between 1 .O and 2.0 nm, a dimension considerable smaller than the 15-nm gold particle diameter. We believe that the contracted dimension in the z direction is a result of operating the microscope in the constant height mode rather than a flattening of the gold colloid particle on the graphite surface. Hence dimensions measured by tunneling are more reliable in the xy than in the z direction, and we conclude that the STM can be used as a reliable and semiquantitativetool for measuring the shape and dimensions of colloidal metallic particles deposited on conducting substrates. After our initial characterization of single, isolated colloidal particles with the STM, we began to explore the properties of colloidal aggregates. The particles were aggregated by adding 1-3 drops of 1 M NaCl or 1 M NaCI04 to 5 cm3 of the colloidal suspension. After the addition of the salt solutions, the wine-red Au suspension turned blue, indicating that the small aggregates of gold were beginning to form.7 At this stage several drops of the suspension were deposited onto a freshly cleaved HOPG substrate and evacuated to dryness. The water was removed at a reduced pressure of Torr without allowing the sol to solidify and sublime. As described earlier, we modified the graphite surface with I-propanol and deposited the aggregates both at ambient temperatures and at 1 IO "C. Elevated temperatures did not change the structure or distribution of aggregates but did speed up the removal of solvent and propanol. Figure 2 shows typical STM and TEM images of colloidal gold aggregates flocculated under these conditions. Samples for TEM analysis were obtained by depositing the colloids on carbon-coated copper grids. The STM images of the flocculated colloid are very similar to the appearance of the aggregates by TEM: the individual particles appear to be physically touching but rarely fused together. As with the isolated particles the aggregates were stable for periods of several hours, and they could be imaged under a wide variety of bias conditions with little effect on their appearance. One of the most important issues in colloid science concerns the nature of the electric double layer which provides the repulsive Coulombic barrier that keeps the particles in suspension. We hoped to use the STM to study the local character of the double layer by aggregating the colloids with different flocculatingagents. For example, ionic salts are known to induce aggregation by causing the collapse of the diffuse double layer surrounding each colloidal particle, while molkular adsorbates are thought to cause aggregation by displacing more weakly bound charged species from the colloid surface. These different agents should produce colloidal surfaces with significantly different electrical and molecular characteristics. Though flocculating with ionic salts produced aggregates with excellent tunneling characteristics, we were not as successful with the molecular flocculating agent, tetrathiafulvalene (TTF).'* In fact we never obtained stable STM images from colloids flocculated with TTF,even though TEM micrographs show that these aggregates are identical with those aggregated with salt. We speculate that adsorbed TTF forms an insulating barrier around the colloid surface that prevents tunneling from occurring. Thus, to obtain the best STM images of flocculated colloids, it seems necessary to aggregate with ionic agents which screen the Cou-
-
50
nm Figure 1. STM image of isolated colloidal gold particles on a HOW surface. The average size of the colloids is I5 nm. The spiky noise in the image is believed to come from salts which precipitate on the graphite surfacc during thc deposition of the colloids.
particularly relevant in the science and technology of dispersed systems. In this Lcttcr we report some initial STM results from colloidal gold deposited onto HOPG. For a number of reasons, the gold system is an ideal starting point for determining the utility of the STM in colloid science. Not only do simple chemical syntheses produce small particles with a narrow size distribution? but both rhe aggregation behavior and surface chemistry of the gold particles have been explored in considerable detail.s-'O Our aqueous gold colloids were prepared through the chemical reduction of sodium tetrachloroaurate using trisodium citrate according to a modified synthesiss first reported by Hillier et This recipe results in the formation of stable gold particles with a mean diameter of 15 nm. These colloidal suspensions remain dispersed for many months, presumably stabilized by citrate ions strongly chemisorbed to the colloid surface. Because the aqueous gold suspension does not wet the graphite, we found it necessary to modify the HOPG surface in order to obtain a sample surface which was uniformly coated with isolated, unaggregated colloidal particles. This modification was achieved by placing a small drop of I-propanol on the graphite surface prior to depositing the gold sol. We found that surfaces treated with the alcohol could be partially dried under vacuum with no apparent loss in wetting efficiency. All of our tunneling images were obtained with an STM operating in air at room temperature." To isolate the system from low-frequency vibrations, the sample head was placed on a 50-kg anvil suspended by rubber cords stretched to approximately twice their equilibrium length. We used Pt/lr STM tips, and operated the microscope at biases of -500 mV to +500 mV and tunneling currents of 1 nA in the constant height mode. To ensure that the STM was operating optimally, we would first Scan the graphite surface under conditions which allowed us to resolve atomic features on the graphite surface. We would then scan over the surface in search of gold particles. Figure I displays the typical appearance of a colloid-covered HOPG surface scanned by the STM under these conditions. The scan reveals a high density of particles on a flat HOPG background. The stability and brightness of the STM images indicate
-
-
(7) Hillier. J.; Turkevitch, J.; Stevenson, P. C. Trum. Furuduy Soc. 1951, 55.
(8) Weitz. D. A.; Lin. M. Y.; Sandroff, C. J. Surf. Sci. 1985, 158, 147. The concentration of sodium citrate used in the synthesis of the gold colloids is incorrect in this reference. It should be IO-*M not IO-' M. (9) Creighton. J. A.; Blatchford. C. G.; Albrecht, M. G. J . Chem. Soc., Faraday Trans. 2 1979, 75,790. (IO) Sandroff, C. J.; Herschbach. D. R. Langmuir 1985. I , 13 1. ( I I ) Nanoscope II. Digital Instruments, Inc., 6780 Cortona Drive, Santa Barbara. CA 931 17.
( 1 2) Deckman, H. W.; Dunsmuir, J. H. J . Vuc.Sci. Technol. B 1983, I , 1109.
The Journal of Physical Chemistry, Vol. 95, No. 2, 1991 505
Letters
Figure 3.
--
STM image of anomalous long-rangc ordcr on graphite. The
periodicity of the hexagonal array was 7.5 nm and it extended over an area of -I50 X 150 nm2. The arrays were always observed near a terrace like the one seen in the lower part of the figure.
F
20 nm Figure 2. S T M and T E M images of colloidal gold aggregates: (a, top) STM image on graphite; (b, bottom) TEM image on carbon-coated copper grid. The colloid sizes were found t o be 15 nm by both techniqucs.
-
lombic interaction between particles without hindering the tunneling process. While exploring the properties of colloidal aggregates we often noticed that the STM images contained sharp spikes. Believing these were due to the presence of codeposited salts, we treated the aggregated suspensions with ion-exchange resin13 to reduce the concentration of ions in solution. Figure 3 displays a remarkable STM image obtained after the aggregated colloids were treated with ion-exchange resin. This image consists of roughly 50 "atoms" which have assumed a highly ordered hexagonal array that measures I50 nm2 X 150 nm2. Each of the "atoms" in the lattice has a diameter of 14 nm, and a two-dimensional Fourier transform identified the lattice constant to be -24 nm. Since previous work had demonstrated that charged colloidal particles deposited from solution onto smooth surfaces can sometimes crystallize in hexagonal lattices,14we originally associated these STM images with our sols, with each "atom" identified as a colloidal gold particle. However, several characteristics of our arrays argue against their being lattices of colloidal gold particles. First, the individual particles are extremely uniform
-
in size, much more uniform than isolated particles appear when seen by either TEM or STM. Also the particles in the lattice do not conduct as much current as isolated gold particles do under identical experimental conditions. Finally, images qualitatively similar to those in Figure 3 were obtained after the deposition of colloidal dispersions which were not treated with ion-exchange resin. Under these circumstancesthe ionic strength of the sol is high, and the long-ranged Coulombic forces necessary to induce two-dimensional crystallization should not be present. We speculate that this highly regular lattice is associated with the graphite substrate itself. Indeed similar images have been obtained from graphite surfaces after the deposition of organic salts from solutions which contained no dispersed particle^.'^ Moreover, very recent studies of graphite surfaces by STM have appeared, and almost identical images were obtained where no deliberate adsorption of material occurred.I6 Kuwabara et a1.I6 argue that the anomalous long-range order seen on graphite can be attributed to Moir6 interference between two slightly misoriented layers of graphite. However, it is possible that several mechanisms are operating simultaneously to give rise to the long-range order on graphite. Indeed, preliminary computer simulations show that anomalous long-range order can develop when a graphite flake picked up by the tip slides across the graphite substrate. I In conclusion, we have shown that the STM can be used with some confidence in the study of the properties of metal colloids. The physical dimensions of colloidal particles can be measured with reasonable accuracy in the plane of the substrate, and large aggregates can be readily images so long as electrically insulating molecules are not adsorbed on the colloid surface. Under some conditions it is possible to observe highly ordered hexagonal arrays on graphite surfaces. Although the origin of this array is not well understood, we believe it is a phenomenon associated with the graphite surface itself. Since the lattice parameter of the ordered array is comparable to the size of colloidal particles, it is important to exercise care when using the STM to study colloids or other macromolecular species adsorbed on graphite surfaces.
'
Acknowledgment. We thank R. S. Robinson for helpful criticism. This research was supported in part by the National Science Foundation under grant CHE-8912674 and by a grant by the Air Force Office of Scientific Research. ~~
( I 3) Sandroff, C. J.; Weitz, D. A.; Chung, J. C.; Herschbach, D. R.; J . Phys. Chem. 1!483,87,2127. (14) We found that unaggregated sol kept over ion-exchange resin ( h e x
MR-3C dry mesh 16-45, Dow Chemical Co.) for 30 min or aggregated sols passed through a 25-mL resin-packed buret gave the best results.
( I 5) Lyding, J. W.; Hubacek, J. S.; Gammie, G.; Skala, S.; Brockenbrough, R.; Shapley, J. R.; Keyes, M. P. J. Vuc. Sci. Technol. 1988, A6.363. (16) Kuwabara, M.; Clarke, D. R.; Smith, D. A. Appl. Phys. Leu. 1990, 56, 2396. ( 17) Womelsdorf, J. F.; Sawamura, M.; Ermler, W. C. Sur/. Sci. Leu., to be published.
